On Tuesday 14 August 2012 10:28:07 Antonio Huete Jimenez wrote: > Folks, > > Patch to skip redo recovery is here: > http://leaf.dragonflybsd.org/~tuxillo/archive/patches/hammer_skipredo02.pat >ch > > I've built an i386 kernel + modules with that patch, and it can be > downloaded from here: > http://leaf.dragonflybsd.org/~tuxillo/archive/temp/kernel_skip_redo.tar.bz2 > > Pierre, patch above adds a tunable that can be used to skip redo recovery > during a normal hammer mount. You need to set vfs.hammer.skip_redo to what > you actually need in your /boot/loader.conf. Possible values are: > > 0 - Run redo recovery normally and fail to mount if > the operation fails (default). > 1 - Run redo recovery, but don't fail to mount if the > operation fails. > 2 - Completely skip redo recovery (only for severe error > conditions and/or debugging. > > You may want to start setting it to 1 and if it doesn't mount, then set it > to 2. > Let us know how it goes.
With it set to 1, it boots, but there are a few messages after it says recovery is optional. With it set to 2, it still boots. Either way, the network interfaces are hosed. When running "netif start", it says "ifconfig: interface *** does not exist", where the three stars are spade, heart, spade. dmesg, whose beginning is cut off, shows that that's ^F^C^F. Various other interfaces whose names are control characters are brought up, none with an IP address, but with lladdr 0, 0.0, 0.0.0, 0.0.0.0, 0.0.0.0.0, and 0.0.0.0.0.0. I set skip_redo to 1, trying to see the errors, and they aren't there. I guess unmounting the filesystem fixed the errors. I booted the old kernel, after doing all this, and succeeded. ifconfig looks normal, except that em0's IPv6 address is missing. I am therefore renaming the previous kernel to kernel, and will keep this one as kernel.skip_redo. I recommend keeping the patch in case this comes up again. Pierre -- gau do li'i co'e kei do
